sbcl-cvs-import Log


Commit Date  
[545fde] by Nikodemus Siivola Nikodemus Siivola

1.0.43.35: fix make-host-2.lisp from 1.0.43.34

THAT was not supposed to go in. Grr. Sorry.

2010-10-07 16:49:30 Tree
[975352] by Nikodemus Siivola Nikodemus Siivola

1.0.43.34: differentiate cross-compiler output from target and host

No difference in the end-product, but seeing "x-compiling" in
build-logs makes them easier to read for slow people like me.

That is:

* while building the xc-host messages are from the host compiler.
If the host happens to be SBCL, that means:

; compiling (DEFUN FOO ...)

* while building the target:

; x-compiling (DEFUN FOO ...)

* while building CLOS and contribs on target:

; compiling (DEFUN FOO ...)

2010-10-07 16:40:47 Tree
[b50d5a] by Alastair Bridgewater Alastair Bridgewater

1.0.43.33: ppc: Implement %ARRAY-ATOMIC-INCF/WORD

* Implement new VOP ARRAY-ATOMIC-INCF/WORD

* Add ppc to the appropriate reader coditionals to
enable use of the new VOP.

2010-10-07 16:37:10 Tree
[ec42c1] by Alastair Bridgewater Alastair Bridgewater

1.0.43.32: x86: Implement %ARRAY-ATOMIC-INCF/WORD.

* Implement new VOP ARRAY-ATOMIC-INCF/WORD

* Add x86 to the appropriate reader conditionals to
enable use of the new VOP.

2010-10-07 16:36:54 Tree
[74f749] by Alastair Bridgewater Alastair Bridgewater

1.0.43.31: x86-64: Implement %ARRAY-ATOMIC-INCF/WORD.

* Implement new VOP ARRAY-ATOMIC-INCF/WORD

* Add x86-64 to the appropriate reader conditionals to
enable use of the new VOP.

2010-10-07 16:36:37 Tree
[ba3bf8] by Alastair Bridgewater Alastair Bridgewater

1.0.43.30: threads: Initial implementation of ATOMIC-INCF for arrays.

* Teach SB-IMPL::EXPAND-ATOMIC-FROB to deal with unboxed
vectors of words as places.

* Define SB-KERNEL:%ARRAY-ATOMIC-INCF/WORD by analogy to
SB-KERNEL:%RAW-INSTACE-ATOMIC-INCF/WORD.

* Add SB-KERNEL:%ARRAY-ATOMIC-INCF/WORD to the VM fndb.

* While we're here, define an interpreter stub for
%ARRAY-ATOMIC-INCF/WORD. %RAW-INSTANCE-ATOMIC-INCF/WORD
still needs one, but that's out of scope right now.

* Note that this is just the arch-independent parts of
ATOMIC-INCF for arrays, the per-arch parts will be separate
commits.

2010-10-07 16:35:53 Tree
[0a15b6] by Nikodemus Siivola Nikodemus Siivola

1.0.43.29: fix OVERAGER-CHARACTER-BUFFERING test-case

* It should be OVEREAGER-CHARACTER-BUFFERING.

* mktemp doesn't allow a suffix after the Xs. Most incidiously, on
Linux it did not even signal an error, but returned "BCL-fifo-XXXXXXX",
causing mkfifo to fail.

2010-10-07 14:10:40 Tree
[40c370] by Nikodemus Siivola Nikodemus Siivola

1.0.43.28: DESCRIBE ALWAYS-BOUND declarations for symbols

EOM

2010-10-06 15:50:34 Tree
[02bc8a] by Nikodemus Siivola Nikodemus Siivola

1.0.43.27: DESCRIBE optimization policy qualities for symbols

They should be in the manual too, but this is a start.

2010-10-06 12:48:17 Tree
[4bc105] by Nikodemus Siivola Nikodemus Siivola

1.0.43.26: propagate-local-call-args for lambdas with optional-dispatches too

Previously we elided the propagation if the lambda had an entry-fun
or an optional-dispatch. The comment notes that we "If the function
has an XEP, then we don't do anything".

There are, however, lambdas with optional-dispatche that don't have
XEPs. Doing propagation for these is required for proper &REST list
type derivation.

Fixes lp#655203.

2010-10-06 08:59:32 Tree
[f59b6f] by Nikodemus Siivola Nikodemus Siivola

1.0.43.25: (LOOP WITH NIL = ...) caused unused variable style-warnings

Patch by Roman Marynchak. Fixes lp#613871.

Always declare #:LOOP-IGNORE variables ignored -- even if they
have initializations.

2010-10-05 08:26:15 Tree
[c99016] by Nikodemus Siivola Nikodemus Siivola

1.0.43.24: initialize cold-layouts with source-location NIL, not 0

Fixes lp#458015.

2010-10-05 07:46:57 Tree
[8f51ee] by Nikodemus Siivola Nikodemus Siivola

1.0.43.23: enable let-conversion for open-coded ALIEN-FUNCALL calls

Otherwise DEBUG > SPEED prevents let-conversion, and leads to
%SAP-ALIEN being left in the code.

Fixes lp#654485.

2010-10-04 11:26:10 Tree
[eb65b7] by Nikodemus Siivola Nikodemus Siivola

1.0.43.22: better errors for invalid :EXTERNAL-FORMAT arguments

Affects OPEN & RUN-PROGRAM. String <-> octets conversions did
the right thing already.

2010-10-04 10:43:39 Tree
[35d1b3] by Nikodemus Siivola Nikodemus Siivola

1.0.43.21: typo in COPYING file

Spotted by Jean-Philippe Paradis.

2010-10-04 09:51:59 Tree
[7bebcb] by Nikodemus Siivola Nikodemus Siivola

1.0.43.20: missing NEWS for 1.0.43.19

EOM

2010-10-04 09:31:02 Tree
[95d168] by Nikodemus Siivola Nikodemus Siivola

1.0.43.19: don't record source-paths for sub-parts of quoted constants

Fixes the performance-half of lp#654289.

2010-10-04 09:27:56 Tree
[29be1c] by Nikodemus Siivola Nikodemus Siivola

1.0.43.18: index SB-EXT:*EVALUATOR-MODE* under #'EVAL in the manual

Looking for "Interpreter" in the concept index isn't probably the
first thing most people do.

Fixes the documentation half of lp#654289.

2010-10-04 09:15:32 Tree
[27b74e] by Nathan Froyd Nathan Froyd

1.0.43.17: fix static symbols for array dispatch tables

The array dispatch tables have gone through several renamings. Unfortunately,
the references to the names in compiler/generic/parms.lisp have not been
renamed as well. Do so, and micro-optimize generic array access slightly.

(Optimizing SYMBOL-VALUE on static/global symbols would help as well...)

2010-10-04 02:12:53 Tree
[425dd2] by Alastair Bridgewater Alastair Bridgewater

1.0.43.16: compiler: Fix non-unicode build.

* Building #-sb-unicode has been broken since 1.0.36.15, due
to a bug in the then-new element type handling for unions of
array types.

* The value originally selected as a sentinel value for not
having processed any of the types in a union was *empty-type*,
which is also the element-type of (array nil (*)), also known
as a subtype of string.

* Simple-string is a union type of (array nil (*)), (array
character (*)), and simple-base-string on sb-unicode targets.
It is a union type of (array nil (*)) and simple-base-string
on non-unicode targets.

* Because the (array nil (*)) came first in the list of
types in the union, and because its element-type was
*empty-type*, the sentinel value, it was ignored when
computing the overall array element type.

* Because the character and base-char types are disjoint,
the overall array element type calculation came up with the
correct answer on unicode builds.

* To correct the problem, select a sentinel value that is
not a type object: NIL.

* From IRC, this morning:
[9:40] * nikodemus hates (array nil)
[9:40] <nikodemus> and it hates me right back

2010-10-03 14:50:53 Tree
[4ad987] by Nikodemus Siivola Nikodemus Siivola

1.0.43.15: use gcc-3 compiler in Cygwin for runtime too

GCC-4.x can not compile with -mno-cygwin; running gcc -mno-cygwin prints
a message advising to use mingw cross-compiler.

Patch by Kalyanov Dmitry.

2010-10-03 08:42:37 Tree
[9b34c5] by Nikodemus Siivola Nikodemus Siivola

1.0.43.14: typo in asdf-module.mk

* gcc-3, not gcc=3. I wonder how I managed to do that.

2010-10-03 08:37:48 Tree
[0d67e1] by Nikodemus Siivola Nikodemus Siivola

1.0.43.13: "minor fixed for Win32" from Kalyanov Dmitry

* fix run-sbcl.sh for cygwin: need to convert the path.

* fix contrib building for cygwin with GCC 4.x installed: require GCC
3.x since GCC 4.x apparently doesn't do -mno-cygwin.

* PeekConsoleInput's third argument is the number of array in elements,
not bytes. http://msdn.microsoft.com/en-us/library/ms684344%28VS.85%29.aspx
Old usage led to stack overwriting.

2010-09-30 08:43:55 Tree
[ded7da] by Nikodemus Siivola Nikodemus Siivola

1.0.43.12: edit STYLE for stricter patch submission guidelines

* "git format-patch -1" is superior to plain diffs, since it includes the commit
message as well -- and the patch author is the best person to write it.

* Explain when to use Launchpad and when to use sbcl-devel.

2010-09-30 08:40:35 Tree
[85e714] by Nikodemus Siivola Nikodemus Siivola

1.0.43.11: smarter timer expiry

When expiring timers, run all expired timers instead of setting the
system timer again after expiring a single one.

Hopefully addresses lp#375515.

2010-09-30 08:39:39 Tree
Older >